Professional Documents
Culture Documents
s.t.
(Fortune)
x1 , x2 , . . . , xn 0
c = (c1 , c2 , . . . , cn )t
(x1 , x2 , . . . , xn )t ,
=
=
(optimization)
,
(b1 , b2 , . . . , bm )
, A = (aij )mn
:
Minimize ct x
(x1 , x2 , . . . , xn )
s.t.
z(x1, x2 , . . . , xn ) =
(P)
c1 x1 + c2 x2 + . . . + cn xn (
Ax b
) ,
x 0
m ()
,
1, 2, m)
, ,
2 82 3
x1 , x2 , . . . , xn ,
Ax b Ax = bAx
(optimality conditions)
b ,
, L. V. Kantorovich
(P)
I.
,
G. B. Dantzig
1947
,
, ,
Kantorovich
, Dantzig
1950 1960 ,
, ,
1975 ,
, Dantzig
L. V. Kantorovich T. C.
Koopmans
1948
(Programming in Lin-
ear Structure)
, T. C. Koop-
( 1979 ),
(Santa Monica)
L. G. Khachian,
1947 1949 ,
N. Z. Shor,D. B. Yudin,
A. S. Nemirovskii (ellip-
(Simplex method),
soid method)
,Khachian
1826
1939 ,W.
Karush
(Linear Programming)3
? 1984
N. Karmarkar
,
,
,
,
II.
,
,
:(1)
(2)
Minimize x1 2x2
s.t.
x1 + x2 40
(P),
2x1 + x2 60
x1 0,
x2 0
:
1:F 6= , F
(feasible do
main) F ={(x1 , x2 ) R2 | x1 +
x2
0,x2
2:F 6= , (P)
0}, P
, F
40, 2x1 + x2
60,x1
(P)
,F 6= ,
(bounded)
, x = (0, 40)
(polytope), (un-
z = 0
80 = 80
,F
2
4 82 3
(P) , n,
, F
n (hyperplane)
(P) (m + n)
1 2, F 6=
( Ax = bx = 0), F
(P) ,F
C(m + n, n) = (m + n)!/n!m!
n = 100,m = 50
C(m + n, n)F
, ,
():
: F
: cT x
III.
,
, F ,
,
.
, F
x2 = 40x1 = 0,x2 = 0
, C(4, 2) = 6
x1 + x2 = 40x2 = 0
(Linear Programming)5
(P) (P)
Big-M Two-Phase
(Fundamental Matrix)
: F x1, Ax1
b,x1 0 , F
= (P) k = 1
k
: x
dkq
(Minimum Ratio Test)
( 2)
F
, ,
(i) dkq ,
, xk (P)
(ii) dkq ,
,
: dkq
F (k )
(i) k , dkq
, (P)
,
nm
, ,
m0.9522 n0.3109 ,
(ii) k , dkq
G. Minty V. Klee
1971
, n
k+1
=x +
k dkq
, 2n
1 2n
k k + 1,
n ,
6 82 3
n =
10002 ,
nm,
,
(P) (D) ,
(Optimality conditions) :
Ax
(polynomial-
b,
x0
A w c,
time)
w0
cT x bt w = 0
()
Rn w
R ,
x (P) ,w (D)
(P) ,
(A, b, c)
(P) (D)
(Dual problem):
Maximize bT w
Mu , Mp q
s.t.
Awc
(D)
w0
,u R , R ,
3 ,
Mu
(D) m
S ={u Rp |Mu }
w1, w2 , . . . , wm n
S 6= , u1 ,
(P) (D)
E 1 , S E 1 6=
(A, b, c),
Mu1
, u1 S E 1
3:
, S E 1 ,
, S E 1
, S E 1 ( 21 E 1
, ,
) 12 E 1 ,
ct x = bt w
E 2,
1 1
E
2
E 2
(Linear Programming)7
u2
Mu , S = ,
S E 1 E 2 , Mu2
, u2 S E 1
u2 E 2 ,
S E 1 21 E 2
12 E 2 E 3 , 21 E 2
E 3 u3
Mu3
3
,
:
: ()
(x, w) Rn+m ,
u1 = (x1 , w1 ),
22L E 1
k = 1
: uk = (xk , wk )
,
xk (P) ,wk (D)
,
E k+1
uk+1
: V (E k+1 ) = 0,uk+1
, (P) (D)
k k + 1,
p,
k
V (E )
V (E
k+1
, V (E k+1) <
1
e 2 (p+1) V (E k ) V (E k+1 )
<
e 2 (p+1) V (E 1 ) e 2 (p+1)
, V (E )k
4 :
8 82 3
: xk (P)
, x = xk
dkx,
k > 0
: xk
xk+1 = xk + k dkx F 0
4
k k + 1,
F ,
,
Big-M Two-
Phase
dkx , k ,
(P) :
Minimize ct x
s.t.
(P)
, ,
Ax = b
x0
Rn
, (P)
, xk
, r > 0,
:
Minimize ct x
b,x > 0} ,
s.t.
F 0 6= ,
(P1 )
n
X
(xj xkj )2 r 2
j=1
1
: x F
k = 1
xkj (j = 1, 2, n)xk j
(Linear Programming)9
(P1 ) , xk ,
, c,
,
x = xk+1 = xk + r
Dk =
k ck
(P1 ) ,
s.t.
(P2 )
x0
..
xkn
Dk1 =
Minimize c x
xk2
Rn+:
t
xk1
1
xk1
1
xk2
..
1
xkn
:
Tk : Rn+ Rn+
x y = Tk (x)=Dk1 x
x > 0,
? xk
xkj ,j = 1, 2, n
,
,
xk yk =
Tk (xk ) = Dk1 xk = e
Tk1 : Rn+ Rn+
y x = Tk1 (y) = Dk y
Tk Rn+ Rn+ , xk
e = (1, 1, 1)t
e
(approximation)
, (P1 )
5
Tk1(e) = Dk e = xk
,(P2)
y:
Minimize ct Dk y
(P2 )
s.t.
n
X
(yj 1)2 1
j=1
(P1 ) , y
Dk c
kDk ck
Dk c
= e
kDk ck
y = yk+1 = yk +
5
xk > 0, n n
10 82 3
x (P2 )
= xk
D k P k Dk c
kPk Dk ck
x = xk+1 = Dk yk+1
Dk2 c
kDk ck
Dk2 c
= xk
kDk ck
= Dk e
(P2 ) ,
,
xk+1 (P)
4:j,j = 1, 2, ,
(P)
n, xk+1
= 0, xk+1 (P)
j
, y (P)
ing :
Minimize ct Dk y
: x1 F 0
s.t.
ADk y = b
n
X
(P )
(yj 1)2 1
j=1
(P2 ) ,(P )
ADk y = b
, (P2 )
Dk c(ADk )
(null space)
k = 1
: dkx = Dk Pk Dk c,
1
,
k =
kPk Dk ck
xk+1 = xk + k dkx
: xk+1
=0,j=
j
1, 2, , n
x = xk+1
k k + 1,
Pk (Dk c)Pk
(P )
y =y
k+1
Pk (Dk c)
=e+
kPk Dk ck
(P)
k+1
x =x
= Dk y
k+1
,
1984 N. Karmarkar
, ,
(Linear Programming)11
IV.
, ,
,
, ,
,
,
,
,
,,
?
,
,
,
,
,
,
, ,
,
, ,
1. Bazaraa, M. S., Jarvis, J. J., and Sherali, H. D., Linear Programming and
Network Flows(second edition), John
Wiley, New York (1990).
2. Bland, R. G., Goldfarb, D., and Todd,
M. J. The ellipsoid method: a survey, Operations Research 29, 10391091 (1981).
3. Borgwardt, K. H., The Simplex
Method:
A probabilistic Analysis,
Springer-Verlag, Berlin (1987).
4. Chvatal, V., Linear Programming,
Freeman, San Francisco (1983).
5. Dantzig, G. B., Maximization of a linear function of variables subject to linear inequalities, Activity Analysis of
Production and Allocation, edited by T.
C. Koopmans, John Wiley and Sons,
New York, 339-347 (1951).
6. Dantzig, G. B., Linear Programming
and Extensions, Princeton University
Press, Princeton, New Jersey (1963).
7. Fang, S. C., and Puthenpura, S.
C., Linear Optimization and Extensions: Theory and Algorithms, Prentice Hall, New Jersey (1993).
8. Gass, S. I., Linear Programming:
Methods and Applications ( second
edition ), McGraw-Hill, New York
(1964).
9. Gilmore, P. C., and Gomory, R. E.,
A linear programming approach
to
the cutting-stoch problem,
Operations Research 9, 849-859 (1961).
12 82 3
14. Karmarkar, N., A new polynomial time algorithm for linear programming, Combinatorica 4, 373-395
(1984).
15. Karush, W., Minima of functions of several variables with
inequalities as side constraints,
Master
thesis,
Department
of
Mathematics, University of Chicago,
Chicago, Illinois (1939).
16. Khachian, L. G., A polynomial algorithm in linear programming, (in Russian) Doklady Akademiia Nauk SSSR
224, 1093-1096, (English translation)
Soviet Mathematics Doklady 20, 191194 (1979). translation), Soviet Mathematics Doklady 6, 286-290 (1965).
,